Using Open Source LLMs

Leverage community-supported Generative AI models in your technology stack.

Pillar

Technology – Platforms, Tools, Infrastructure & Productivity

Overview

This course introduces open-source large language models (LLMs) such as LLaMA, Mistral, and Falcon, exploring how to integrate and customize these models for enterprise applications. Participants will gain practical skills to deploy, fine-tune, and maintain open-source GenAI solutions that offer flexibility and control.

Learning Objectives

Participants will be able to:

• Understand the architecture and capabilities of popular open-source LLMs

• Deploy open-source models within cloud or on-premise environments

• Fine-tune and customize models for specific business needs

• Manage infrastructure and resource considerations for LLM workloads

• Evaluate performance and scalability of open-source GenAI solutions

Target Audience

• AI engineers and data scientists

• Machine learning operations (MLOps) professionals

• Cloud architects and infrastructure engineers

• Technical leads overseeing AI deployments

Duration

20 hours over 4 days (5 hours per day)

Delivery Format

• Hands-on deployment labs and model customization sessions

• Technical deep dives and architecture discussions

• Collaborative troubleshooting and optimization exercises

Materials Provided

• Deployment guides and scripts for popular open-source LLMs

• Sample datasets and fine-tuning configurations

• Access to test environments for practice

• Certificate of completion

Outcomes

• Confidently deploy and manage open-source LLMs in production

• Customize models to better fit domain-specific tasks

• Optimize infrastructure for cost and performance efficiency

• Maintain and update GenAI models responsibly and securely

Outline / Content

Day 1: Introduction to Open Source LLMs

• Overview of LLaMA, Mistral, Falcon, and others

• Comparing architectures and use cases

• Licensing and community support considerations

Day 2: Deployment and Integration

• Setting up models on cloud and on-premise environments

• Managing dependencies and infrastructure requirements

• Integrating models with business applications via APIs

Day 3: Fine-Tuning and Customization

• Preparing datasets for domain adaptation

• Techniques for fine-tuning and transfer learning

• Evaluating model performance post customization

Day 4: Operations and Best Practices

• Monitoring model health and performance

• Scaling models to meet demand

• Security and ethical considerations in open-source GenAI

• Workshop: Deploy and fine-tune an open-source LLM for a sample use case